From ce0e72877a72fce943800908fe7ad83d0b8e9415 Mon Sep 17 00:00:00 2001
From: Nan Xiao <me@nanx.me>
Date: Fri, 26 Apr 2024 23:19:23 -0400
Subject: [PATCH] Add code examples for wrapper functions

---
 R/stackgbm.R           | 23 ++++++++++++-----------
 R/wrappers_lightgbm.R  | 28 +++++++++++++++++++++++++++-
 R/wrappers_xgboost.R   | 41 +++++++++++++++++++++++++++++++++++++++--
 man/lightgbm_train.Rd  | 28 +++++++++++++++++++++++++++-
 man/xgboost_dmatrix.Rd | 15 ++++++++++++++-
 man/xgboost_train.Rd   | 26 +++++++++++++++++++++++++-
 6 files changed, 144 insertions(+), 17 deletions(-)

diff --git a/R/stackgbm.R b/R/stackgbm.R
index b6679ae..ff25120 100644
--- a/R/stackgbm.R
+++ b/R/stackgbm.R
@@ -35,7 +35,7 @@ stackgbm <- function(x, y, params, n_folds = 5L, seed = 42, verbose = TRUE) {
   x_glm <- matrix(NA, nrow = nrow_x, ncol = 3L)
   colnames(x_glm) <- c("xgb", "lgb", "cat")
 
-  # xgboost
+  # xgboost ----
   pb <- progress_bar$new(
     format = "  fitting xgboost model [:bar] :percent in :elapsed",
     total = n_folds, clear = FALSE, width = 60
@@ -68,7 +68,7 @@ stackgbm <- function(x, y, params, n_folds = 5L, seed = 42, verbose = TRUE) {
     x_glm[index_xgb == i, "xgb"] <- predict(fit, xtest)
   }
 
-  # lightgbm
+  # lightgbm ----
   pb <- progress_bar$new(
     format = "  fitting lightgbm model [:bar] :percent in :elapsed",
     total = n_folds, clear = FALSE, width = 60
@@ -100,7 +100,7 @@ stackgbm <- function(x, y, params, n_folds = 5L, seed = 42, verbose = TRUE) {
     x_glm[index_lgb == i, "lgb"] <- predict(fit, xtest)
   }
 
-  # catboost
+  # catboost ----
   pb <- progress_bar$new(
     format = "  fitting catboost model [:bar] :percent in :elapsed",
     total = n_folds, clear = FALSE, width = 60
@@ -130,17 +130,18 @@ stackgbm <- function(x, y, params, n_folds = 5L, seed = 42, verbose = TRUE) {
     x_glm[index_cat == i, "cat"] <- catboost_predict(fit, pool = test_pool, prediction_type = "Probability")
   }
 
-  # logistic regression
+  # Logistic regression ----
   df <- as.data.frame(cbind(y, x_glm))
   names(df)[1] <- "y"
   model_glm <- glm(y ~ ., data = df, family = binomial())
 
-  lst <- list(
-    "model_xgb" = model_xgb,
-    "model_lgb" = model_lgb,
-    "model_cat" = model_cat,
-    "model_glm" = model_glm
+  structure(
+    list(
+      "model_xgb" = model_xgb,
+      "model_lgb" = model_lgb,
+      "model_cat" = model_cat,
+      "model_glm" = model_glm
+    ),
+    class = "stackgbm"
   )
-  class(lst) <- "stackgbm"
-  lst
 }
diff --git a/R/wrappers_lightgbm.R b/R/wrappers_lightgbm.R
index 9f62c30..f7595b4 100644
--- a/R/wrappers_lightgbm.R
+++ b/R/wrappers_lightgbm.R
@@ -10,7 +10,33 @@
 #' @export
 #'
 #' @examplesIf is_installed_lightgbm()
-#' # Example code
+#' sim_data <- msaenet::msaenet.sim.binomial(
+#'   n = 100,
+#'   p = 10,
+#'   rho = 0.6,
+#'   coef = rnorm(5, mean = 0, sd = 10),
+#'   snr = 1,
+#'   p.train = 0.8,
+#'   seed = 42
+#' )
+#'
+#' fit <- suppressWarnings(
+#'   lightgbm_train(
+#'     data = sim_data$x.tr,
+#'     label = sim_data$y.tr,
+#'     params = list(
+#'       objective = "binary",
+#'       learning_rate = 0.1,
+#'       num_iterations = 100,
+#'       max_depth = 3,
+#'       num_leaves = 2^3 - 1,
+#'       num_threads = 1
+#'     ),
+#'     verbose = -1
+#'   )
+#' )
+#'
+#' fit
 lightgbm_train <- function(data, label, params, ...) {
   rlang::check_installed("lightgbm", reason = "to train the model")
   cl <- rlang::call2(
diff --git a/R/wrappers_xgboost.R b/R/wrappers_xgboost.R
index 7cf4543..548ecfc 100644
--- a/R/wrappers_xgboost.R
+++ b/R/wrappers_xgboost.R
@@ -9,7 +9,20 @@
 #' @export
 #'
 #' @examplesIf is_installed_xgboost()
-#' # Example code
+#' sim_data <- msaenet::msaenet.sim.binomial(
+#'   n = 100,
+#'   p = 10,
+#'   rho = 0.6,
+#'   coef = rnorm(5, mean = 0, sd = 10),
+#'   snr = 1,
+#'   p.train = 0.8,
+#'   seed = 42
+#' )
+#'
+#' x_train <- xgboost_dmatrix(sim_data$x.tr, label = sim_data$y.tr)
+#' x_train
+#' x_test <- xgboost_dmatrix(sim_data$x.te)
+#' x_test
 xgboost_dmatrix <- function(data, label = NULL, ...) {
   rlang::check_installed("xgboost", reason = "to create a dataset")
   cl <- if (is.null(label)) {
@@ -32,7 +45,31 @@ xgboost_dmatrix <- function(data, label = NULL, ...) {
 #' @export
 #'
 #' @examplesIf is_installed_xgboost()
-#' # Example code
+#' sim_data <- msaenet::msaenet.sim.binomial(
+#'   n = 100,
+#'   p = 10,
+#'   rho = 0.6,
+#'   coef = rnorm(5, mean = 0, sd = 10),
+#'   snr = 1,
+#'   p.train = 0.8,
+#'   seed = 42
+#' )
+#'
+#' x_train <- xgboost_dmatrix(sim_data$x.tr, label = sim_data$y.tr)
+#'
+#' fit <- xgboost_train(
+#'   params = list(
+#'     objective = "binary:logistic",
+#'     eval_metric = "auc",
+#'     max_depth = 3,
+#'     eta = 0.1
+#'   ),
+#'   data = x_train,
+#'   nrounds = 100,
+#'   nthread = 1
+#' )
+#'
+#' fit
 xgboost_train <- function(params, data, nrounds, ...) {
   rlang::check_installed("xgboost", reason = "to train the model")
   cl <- rlang::call2(
diff --git a/man/lightgbm_train.Rd b/man/lightgbm_train.Rd
index 8bb6cd9..e21c7e4 100644
--- a/man/lightgbm_train.Rd
+++ b/man/lightgbm_train.Rd
@@ -23,6 +23,32 @@ Train lightgbm model
 }
 \examples{
 \dontshow{if (is_installed_lightgbm()) (if (getRversion() >= "3.4") withAutoprint else force)(\{ # examplesIf}
-# Example code
+sim_data <- msaenet::msaenet.sim.binomial(
+  n = 100,
+  p = 10,
+  rho = 0.6,
+  coef = rnorm(5, mean = 0, sd = 10),
+  snr = 1,
+  p.train = 0.8,
+  seed = 42
+)
+
+fit <- suppressWarnings(
+  lightgbm_train(
+    data = sim_data$x.tr,
+    label = sim_data$y.tr,
+    params = list(
+      objective = "binary",
+      learning_rate = 0.1,
+      num_iterations = 100,
+      max_depth = 3,
+      num_leaves = 2^3 - 1,
+      num_threads = 1
+    ),
+    verbose = -1
+  )
+)
+
+fit
 \dontshow{\}) # examplesIf}
 }
diff --git a/man/xgboost_dmatrix.Rd b/man/xgboost_dmatrix.Rd
index 8fd4880..edb1762 100644
--- a/man/xgboost_dmatrix.Rd
+++ b/man/xgboost_dmatrix.Rd
@@ -21,6 +21,19 @@ Create xgb.DMatrix object
 }
 \examples{
 \dontshow{if (is_installed_xgboost()) (if (getRversion() >= "3.4") withAutoprint else force)(\{ # examplesIf}
-# Example code
+sim_data <- msaenet::msaenet.sim.binomial(
+  n = 100,
+  p = 10,
+  rho = 0.6,
+  coef = rnorm(5, mean = 0, sd = 10),
+  snr = 1,
+  p.train = 0.8,
+  seed = 42
+)
+
+x_train <- xgboost_dmatrix(sim_data$x.tr, label = sim_data$y.tr)
+x_train
+x_test <- xgboost_dmatrix(sim_data$x.te)
+x_test
 \dontshow{\}) # examplesIf}
 }
diff --git a/man/xgboost_train.Rd b/man/xgboost_train.Rd
index e3ab51d..69c31c5 100644
--- a/man/xgboost_train.Rd
+++ b/man/xgboost_train.Rd
@@ -23,6 +23,30 @@ Train xgboost model
 }
 \examples{
 \dontshow{if (is_installed_xgboost()) (if (getRversion() >= "3.4") withAutoprint else force)(\{ # examplesIf}
-# Example code
+sim_data <- msaenet::msaenet.sim.binomial(
+  n = 100,
+  p = 10,
+  rho = 0.6,
+  coef = rnorm(5, mean = 0, sd = 10),
+  snr = 1,
+  p.train = 0.8,
+  seed = 42
+)
+
+x_train <- xgboost_dmatrix(sim_data$x.tr, label = sim_data$y.tr)
+
+fit <- xgboost_train(
+  params = list(
+    objective = "binary:logistic",
+    eval_metric = "auc",
+    max_depth = 3,
+    eta = 0.1
+  ),
+  data = x_train,
+  nrounds = 100,
+  nthread = 1
+)
+
+fit
 \dontshow{\}) # examplesIf}
 }